Article Text

Download PDFPDF

Retinopathy of prematurity
  1. S P Donahue
  1. Department of Ophthalmology and Visual Sciences, 8th Floor, Vanderbilt Medical Center East, 1215 21st Avenue, South, Nashville, TN 37232, USA

    Statistics from

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    We should focus our screening efforts to target those most at risk

    Despite major advances in the treatment of retinopathy of prematurity (ROP), this disease remains a significant cause of blindness in the developed world. ROP as we know it probably did not exist before the advance in prenatal care that occurred in the mid-1950s, with enhanced survival that allowed children to live at an age where retinal vascularisation was incomplete and ROP could develop.

    Additional recent advances in neonatology have influenced the incidence and severity of ROP in a dichotomous fashion. While many children who would have fought for survival two decades ago now remain quite healthy throughout their neonatal stay, others for whom death was nearly certain now fight for survival with reasonable odds. Thus, older children are healthier and probably at a decreased risk of ROP, whereas smaller children who would have died now live and are at high risk. How the interplay of these changes has influenced the incidence and severity of ROP is controversial.

    In this issue of the BJO (1122), Larsson et al examine this question. The previous studies they cite were generally small, retrospective, and poorly controlled. Larsson et al, however, were able to utilise the geography of Stockholm, Sweden, to perform a rigorous evaluation of the effects of advances in neonatal care on the incidence and severity of ROP.

    In 1993, these investigators published results from 350 children born weighing 1500 g or less, in a well defined geographical area of Stockholm where it is highly likely that all such live births were identified (1990 cohort). Sixteen per cent of the children died before 7 weeks of age, and follow up data were available for 260 children. Overall, 40.4% developed some ROP, 20% developed stage III ROP, and 10.8% were treated (at a slightly earlier stage than utilised in the CRYO-ROP). The authors compared these data with those obtained in the same manner, from the same location, 10 years later (2000 cohort), virtually eliminating many of the biases found in other comparison studies. In that cohort, 88% of the 331 liveborn infants <1500 g during the study period survived, and they were able to follow 253. ROP developed in 36.4%, and progressed to stage III or greater in 18.2%. Thirty one (12.3%) of these children needed treatment. These percentages were statistically indistinct from those reported a decade earlier. However, the distribution of ROP had shifted so that younger and smaller children accounted for the majority of the cases of ROP.

    I believe the important message of this paper is the major shift in the distribution of ROP; while 16% of children >30 weeks’ gestational age had severe ROP in the 1990 cohort, and half were treated, only two children (1%) of the similar 2000 cohort developed severe ROP, and neither progressed sufficiently to require treatment. The results were similar, but slightly less striking for the 750–1000 g infants as well: 18.4% of the 1990 cohort had at least stage III ROP compared to fewer than 9% of the 2000 cohort. These results were statistically significant by a wide margin.

    Most of the children who develop severe ROP now would have died had they been born a decade ago

    How can one reconcile the virtual elimination of ROP in children over 1000 g with the similar overall incidence data between the cohorts? There are two possible explanations: either very small children are developing ROP at a higher rate now than they did 10 years ago, or some type of bias is involved in the sampling. I believe it to be the latter: most of the children who develop severe ROP now would have died had they been born a decade ago. Children born at 26 weeks or less of gestational age represented fewer than 10% of the 1990 cohort, but 23% of the 2000 group. The rate of severe ROP for such young children in 2000 was 61%, and nearly all of those progressed to require treatment. Assuming similar rates of premature births in the two cohorts, one can calculate that the 2000 cohort includes approximately 23 children of 26 gestational weeks at birth who survived but would have died a decade earlier, and at least 20 of these 23 children developed ROP. If these 23 children are eliminated from the analysis of ROP incidence, the overall incidence of total ROP for the 2000 cohort drops to 72/230 (31%), severe ROP to 14% (32/230), and treatable ROP to 17/230 (7.4%). These numbers are consistent with the approximately 30% decrease in ROP incidence reported in other studies. Thus, the authors’ observation that the overall incidence of ROP is unchanged is an epiphenomenon caused by the tremendous increase in survival rate of infants at 26 weeks of age, all of whom appear to be at high risk of severe disease. This then begs the question of whether Swedish intensive care units (ICUs) are now saving children that American ICUs did a decade ago, or if some other factor either decreases the survival rate, or the ROP rate, of similarly aged premature children in the United States.

    A major result not pointed out by the authors concerns the effect of treatment: in 1990, seven children (2.7%) developed stage 4 or 5 disease. This decreased to only two children in 2000. It is unclear whether this improved result comes from an advantage of laser treatment over CRYO therapy for ROP, an inherent but unconscious bias to treat higher risk children earlier, or a statistical aberration, but it warrants further investigation.

    ROP will continue to be a blinding disease despite further research investigating the pathophysiology of this disease, and new studies (such as ETROP) that help define which children will develop treatable disease. Therefore, it is imperative to determine with more accuracy which children are at highest risk and need to be screened. The authors are to be commended for a rigorous and valid report that helps to focus our screening efforts to target those most at risk.

    Linked Articles